Oncofusion

start / graph / mechanism-atm-ell-t551-s561-s589-eaf1-switch

Mechanismmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q25-partner-phospho

ATM phosphorylates ELL at T551, S561 and S589 - inside the OCEL domain this graph already calls essential for MLL-ELL immortalisation

also called ELL T551 S561 S589 · ATM ELL phosphorylation · ELL EAF1 switch · ELL self-association

When DNA is damaged, the alarm kinase ATM tags the elongation factor ELL at three spots; the tagged ELL grabs its partner EAF1 harder, clumps with itself, and lets go of the machine that drives genes to be read out. All three spots sit inside the exact piece of ELL that the leukaemia fusion keeps and needs.

The same thing, for a scientist

Upon genotoxic stress ATM phosphorylates human ELL at T551, S561 and S589 - three of the seven S/T-Q motifs in P55199 and the only three in the C-terminal region. An alanine triple mutant (T551A, S561A, S589A) shows markedly reduced stress-induced phosphorylation, reduced EAF1 interaction and defective stress-induced self-association. Mechanistically ELL has an intrinsic self-association property that reduces its availability to other SEC components; EAF1 binding enhances that self-association, so ATM phosphorylation shifts ELL out of the SEC and shuts down global transcription. **All three residues lie between 551 and 589, inside the OCEL/occludin-homology domain (aa 507-617)** - which this graph's `subtraction-ell-n-terminal-eaf-site` entry already records as containing the C-terminal EAF1-interaction site that is *critical for MLL-ELL immortalisation*, while the ELL elongation domain is dispensable. So ATM phosphorylation lands on the one part of ELL that the fusion needs, and S561 is independently observed as a phosphosite in PeptideAtlas and in UniProt's curated table.

Go deeper
The switch
  1. ELL self-associates intrinsically, which reduces its interaction with other SEC components 1.
  2. EAF1 binds ELL and enhances that self-association; contrary to in vitro expectation, EAF1 therefore inhibits ELL-dependent Pol II transcription in cells 1.
  3. Genotoxic stress activates ATM, which phosphorylates ELL at T551, S561 and S589 1.
  4. Phospho-ELL binds EAF1 more, self-associates more, and leaves the SEC - global transcriptional inhibition 1.
  5. A T551A/S561A/S589A triple mutant is defective in all of steps 3, 4 and the self-association readout 1.

An earlier high-throughput study had already reported ATM-mediated ELL phosphorylation at S561, which the authors cite; so one of the three residues has independent prior support 1.

Retained, and on the part that matters

MLL-ELL contains ELL residues 46-621; the discarded segment is ELL exon 1, aa 1-45 2. T551, S561 and S589 are all retained, with a wide margin.

More than retained: they are inside the OCEL domain (507-617), and this graph already records that the C-terminal EAF1-interaction site within that region is critical for MLL-ELL immortalisation while the ELL elongation-stimulatory region is dispensable 2. So the ATM sites sit in the module whose EAF1 contact is the transforming element.

That makes a specific and testable prediction: ATM phosphorylation of the MLL-ELL fusion should increase its EAF1 binding, the very interaction the fusion requires. Whether that helps or hurts the fusion is genuinely open - more EAF1 binding is what transformation needs, but the same phosphorylation drives wild-type ELL out of the SEC. The fusion has no SEC-facing N-terminus to lose, so the two arms of the wild-type switch may not both apply.

All twelve observed ELL phosphosites are retained, because the most N-terminal is S126 3.

What has never been done

No experiment anywhere in this beat's searches expresses MLL-ELL and asks whether ATM phosphorylates it. The primary study is entirely wild-type ELL in non-haematopoietic human cell lines with doxorubicin as the stressor, and the word 'leukemia' appears in it only as part of the protein's name.

Facts and figures
kinasegene-atm
kinase nameATM
substrategene-ell
uniprotP55199
motifS/T-Q; three of the seven [ST]Q motifs in P55199 (59, 203, 235, 345, 551, 561, 589)
gated contactEAF1 binding; ELL self-association; loss of SEC interaction
directionphosphorylation INCREASES EAF1 binding and self-association, DECREASES SEC interaction
domain contextall three inside the OCEL / occludin-homology domain, aa 507-617
mapped inwild-type ELL, human cell lines, doxorubicin
mapped in fusionfalse
retained in fusiontrue
retained basisMLL-ELL contains ELL aa 46-621; all three sites are far C-terminal of 45
prior independent reportS561 reported as an ATM site by earlier high-throughput analysis, cited by the primary paper
correction on recordPMID 37194690
gene symbolELL
sites
[
  551,
  561,
  589
]
site residues
{
  "551": "T",
  "561": "S",
  "589": "S"
}
observed in peptideatlas
{
  "551": false,
  "561": true,
  "589": false
}
How this could bite us back
  • ATM inhibitors are in oncology trials. In KMT2A::ELL disease an ATM inhibitor would prevent the phosphorylation that increases the fusion's EAF1 binding - which could be beneficial, or could preserve the fusion's SEC engagement during chemotherapy. The direction is not predictable from the wild-type data and would need measuring before combining.
  • The transforming EAF1 contact and the damage-response EAF1 contact are the same interaction on the same domain. Any molecule designed to block the ELL-EAF1 interface in leukaemia would also block a general DNA-damage-response mechanism in every dividing cell.
  • S561 is the only one of the three residues with independent proteomic observation; T551 and S589 rest on one laboratory's mutant. A phospho-specific reagent programme built on all three would be building two thirds of it on a single triple-mutant experiment.
Still open
  1. Is MLL-ELL phosphorylated at T551/S561/S589 after genotoxic stress, and does that increase its EAF1 binding - the interaction the fusion needs to immortalise?
  2. The wild-type switch has two arms: more EAF1 and less SEC. The fusion lacks ELL's N-terminal EAF site and has the KMT2A half in place of an ELL N-terminus. Do both arms still apply, or only the EAF1 one?
  3. Does a T551A/S561A/S589A MLL-ELL transform myeloid progenitors better, worse, or the same? A single retroviral immortalisation assay with the triple mutant would be decisive and cheap.
  4. Individual contributions of the three residues are unresolved because only the triple mutant was tested.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ATM phosphorylates human ELL at T551, S561 and S589 during genotoxic stress; a T551A/S561A/S589A triple mutant shows significantly reduced ATM-mediated phosphorylation, reduced EAF1 interaction and defective stress-induced self-association, and enhanced EAF1 association reduces ELL interaction with other SEC components, producing global transcriptional inhibition.

    PMID 36305813DOI
    What this evidence cannot show

    Full text read via Europe PMC fullTextXML (PMC9638944) on 2026-09-19. Human cell lines, doxorubicin-induced damage, largely ectopic ELL with endogenous confirmation for parts; no haematopoietic system and no MLL-ELL fusion protein anywhere in the study. The three residues were selected as C-terminal S/T-Q candidates and tested as a triple mutant, so the individual contribution of each is not resolved and residual phosphorylation remains in the triple mutant. This article has a published correction which must be read with it.

    Evidence that argues against this (1)
    • A correction to this article is on record at Nucleic Acids Research; the primary report should not be cited without it. Nucleic Acids Res 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-ELL contains ELL amino acids 46 to 621, discarding only ELL exon 1 (aa 1-45); the retained C-terminal EAF1-interaction domain within/near the OCEL region is critical for immortalisation while the ELL elongation-stimulatory domain is dispensable.

    PMID 11463848DOI
    What this evidence cannot show

    Cited here through this graph's existing subtraction-ell-n-terminal-eaf-site entry, which resolved the 46-621 construct boundary from this paper's full text (PMC87288) in a previous cycle. This session did not re-read that full text, and this is a 2001 retroviral myeloid-progenitor immortalisation assay, not a patient-derived system.

    Evidence that argues against this (1)
    • The same graph entry records that MLL-ELL also delocalises endogenous EAF1 out of nuclear speckles into diffuse nucleoplasm, a dominant effect on wild-type ELL's interactions that is documented but never causally tested. If the fusion is already deranging EAF1 distribution, a phospho-switch that increases EAF1 affinity may operate on a pool that is not where the wild-type switch puts it. Simone F et al. 2003
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    ELL S561 is independently observed as a phosphosite in both the PeptideAtlas/PTMeXchange re-analysis and UniProt's curated Modified-residue table for P55199, and all twelve observed ELL phosphosites lie at residue 126 or higher, hence on sequence retained in MLL-ELL.

    What this evidence cannot show

    Fetched 2026-09-19. Proteomic observation of a residue is not evidence about which kinase produced it, nor about the cell type or condition. T551 and S589 are NOT in the observed set, so two of the three ATM sites have functional but not proteomic support.

    Evidence that argues against this (1)
    • A systematic re-analysis of public phosphoproteomic datasets finds S309, not S561, to be the most frequently detected and most consistently regulated ELL phosphosite, and does not identify the ATM triple-site cluster as a dominant feature of the landscape. Frequency of detection and functional importance are separate quantities and they disagree here. Kammarambath SR et al. 2026
Connected to